Given the uncertainty of the service environment, the quality assessment of scrap parts became more complex. A damage-based multi-state mapping model (DBMS) was proposed herein based on Dirichlet distribution. The model determined the main failure characteristics by analyzing the failure behaviors of the scrap parts, adopted multinomial distribution for mathematical abstraction of the damage data of parts, and selected Dirichlet distribution as the prior probability distribution. The posterior distribution parameters were updated by Bayes formula, and the posterior probability expected value of damage data mapped to different quality levels was obtained. Further, D-S evidence theory was introduced to integrate damage information to realize the comprehensive assessment of the quality of scrap parts. In order to verify the feasibility and effectiveness of the model, waste worm gear was taken as the case study object and compared with existing literature methods. The experimental results show that the model has advantages in prediction accuracy and generalization ability.
D-S证据理论是一种用于处理不确定信息的数学框架,通过组合规则整合来自不同信息源的证据,适用于各种不确定性环境下的决策和推理。辨识框架Θ是D-S证据理论的基础,它表示了问题所有可能结果的完备集合,Θ中的元素数量有限且相互排斥。辨识框架的幂集2 Θ 包含了Θ的所有子集,每个子集对应一个命题。基本概率分配(basic probability assignment,BPA)是赋予幂集内不同命题可能性的映射函数,数学表达式为,且满足以下条件:
在贝叶斯非参数方法中,Dirichlet分布被认定为多项分布的自然共轭先验。引入参数为 a =(a1,a2,…,ak )的Dirichlet分布作为多项分布零部件损伤量数据的先验概率模型,概率向量 θ 被视为在先验基准 a 条件下的随机变量,即P( θ )~D( a )。
在多数实际应用场景中,直接估计先验Dirichlet分布参数 a 常常受限于废旧零部件失效数据的稀缺或信息不足,为降低对先验知识或假设的过度依赖,采用非信息性先验推断方法以得到稳健且有效的先验分布。Jeffreys先验是一种常用的非信息先验[12-13],基于Fisher信息矩阵来进行构造,不依赖于任何特定的参数值。根据文献[14]中的数学推导,在废旧零部件失效损伤量多项分布的情境下,Jeffreys先验遵循以下关系:
GUOHongfei, LUXinyu, RENYaping, et al. A Group Evolutionary Algorithm Based on Reinforcement Learning for Solving Bilateral Multi-objective Simultaneous Parallel Disassembly Line Balance Problem[J]. Chinese Journal of Mechanical Engineering, 2019,59(7):355-366.
CHENGXianfu, ZHOUJian, YOUMinhua, et al. Modular Approach to Product Remanufacturing Based on Failure Mode Transfer Network Model[J]. Computer Integrated Manufacturing Systems, 2023, 29(3):896-909.
SHULinsen, GONGJiangtao, DONGYue, et al. Process Method and Experimental Study of Laser Remanufacturing of Valve Spool Parts[J]. China Mechanical Engineering,2023,34(4):446-453.
GUOHongfei, CHENZhibin, RENYaping, et al. Research on Disassembly Sequence and Disassembly Depth Integration Decision Based on Comprehensive Evaluation of Parts Recovery[J]. Chinese Journal of Mechanical Engineering, 2019,58(4):258-268.
[9]
PONTEB, CANNELLAS, DOMINGUEZR, et al. Quality Grading of Returns and the Dynamics of Remanufacturing[J]. International Journal of Production Economics, 2021, 236(1/2):108129.
[10]
SUNH, CHENW, LIUB, et al. Economic Lot Scheduling Problem in a Remanufacturing System with Returns at Different Quality Grades[J]. Journal of Cleaner Production, 2018, 170:559-569.
[11]
YANIKOĞLUİ, DENIZELM. The Value of Quality Grading in Remanufacturing under Quality Level Uncertainty[J]. International Journal of Production Research, 2020, 59(3):839-859.
[12]
FERGUSONM, GUIDEV D, KOCAE, et al. The Value of Quality Grading in Remanufacturing[J]. Production and Operations Management, 2009, 18(3):300-314.
[13]
ZHANGXugang, WANGYuling, XIANGQin, et al. Remanufacturability Evaluation Method and Application for Used Engineering Machinery Parts Based on Fuzzy-EAHP[J]. Journal of manufacturing systems, 2020, 57:133-147.
[14]
OMWANDOT A, OTIENOW A, FARAHANIS, et al. A Bi-level Fuzzy Analytical Decision Support Tool for Assessing Product Remanufacturability[J]. Journal of Cleaner Production, 2018, 174:1534-1549.
[15]
AKRAMM, ILYASF, DEVECIM. Interval Rough Integrated SWARA-ELECTRE Model:an Application to Machine Tool Remanufacturing[J]. Expert Systems with Applications, 2024, 238:122067.
[16]
SAPUTROD R S, AMALIAF, WIDYANINGSIHP, et al. Parameter Estimation of Multivariate Multiple Regression Model Using Bayesian with Non-informative Jeffreys’ Prior Distribution[J]. Journal of Physics:Conference Series, 2018, 1022(1):012002.
[17]
LIMingming, SUNHuafei, PENGLinyu. Fisher-rao Geometry and Jeffreys Prior for Pareto Distribution[J]. Communications in Stsatistics—Theory and Methods, 2022, 51(6):1895-1910.
GELMANA, CARLINJ B, STERNH S, et al. Bayesian Data Analysis[M]. Boca Raton: CRC Press, 2013.
[21]
WANGHan, JIANGZhigang, ZHANGXugang, et al. A Fault Feature Characterization Based Method for Remanufacturing Process Planning Optimization[J]. Journal of Cleaner Production, 2017, 161:708-719.
[22]
ZHANGXugang, AOXiuyi, JIANGZhigang, et al. A Remanufacturing Cost Prediction Model of Used Parts Considering Failure Characteristics[J]. Robotics and Computer-Integrated Manufacturing, 2019, 59:291-296.
[23]
WANGYanhong, JIANGZhigang, HUXiaoli, et al. Optimization of Reconditioning Scheme for Remanufacturing of Used Parts Based on Failure Characteristics[J]. Robotics and Computer-integrated Manufacturing, 2020, 61:101833.
[24]
JIANGWen, DUANMUDejie, FANXin, et al. A New Method to Determine Basic Probability Assignment under Fuzzy Environment[C]∥2012 International Conference on Systems and Informatics. Yantai, 2012:758-762.
[25]
ZHANGShengjia, XIAOFuyuan. A TFN-based Uncertainty Modeling Method in Complex Evidence Theory for Decision Making[J]. Information Sciences, 2023, 619:193-207.